Post

Hi,
I just installed FramePack Studio. I used the pre-defined installer, which did the job quick and well. Pinokio was installed some days earlier, cause I first did some videos with the "native" FramePack.
As I now started FramePack-Studio, I got this blue-screen message:

--- Attention Configuration ---
⚠️ No attention library found. Using native PyTorch Scaled Dot Product Attention.

• For better performance, consider installing one of:
  SAGE Attention (highest performance), Flash Attention (high performance), or xFormers.

---

Traceback (most recent call last):
File "E:\Programme\pinokio\api\fp-studio.git\app\studio.py", line 33, in
from diffusers_helper.memory import cpu, gpu, get_cuda_free_memory_gb, move_model_to_device_with_memory_preservation, offload_model_from_device_for_memory_preservation, fake_diffusers_current_device, DynamicSwapInstaller, unload_complete_models, load_model_as_complete
File "E:\Programme\pinokio\api\fp-studio.git\app\diffusers_helper\memory.py", line 8, in
gpu = torch.device(f'cuda:{torch.cuda.current_device()}')
File "E:\Programme\pinokio\api\fp-studio.git\app\env\lib\site-packages\torch\cuda_init_.py", line 1148, in current_device
lazy_init()
File "E:\Programme\pinokio\api\fp-studio.git\app\env\lib\site-packages\torch\cuda_init.py", line 471, in _lazy_init
raise AssertionError("Torch not compiled with CUDA enabled")
AssertionError: Torch not compiled with CUDA enabled

(env) (base) E:\Programme\pinokio\api\fp-studio.git\app>
<<<<<<

My System:
Windows 11 home 64x
GPU: RTX 4080 Super

So what to do about "Torch not compiled with CUDA enabled"?
As said: The native FramePack works well using my GPU-Card.

Regards
Andy